Lektra provides on‑demand GPU compute powered by distributed solar and microgrid resources, turning renewable installations into mini data centers. Users can provision H200 or RTX PRO 6000 nodes via a cloud‑native API or dashboard with transparent per‑hour pricing and no egress fees, while the renewable‑fed architecture ensures high availability during grid outages.
Funding
Funding not disclosed
Founders
Product
Problem
Traditional cloud providers rely on centralized, grid‑powered data centers that are costly, carbon‑intensive, and vulnerable to power disruptions, limiting affordable access to high‑performance GPU compute for many AI workloads.
Solution
Lektra builds a distributed, solar‑powered compute fabric by converting rooftop solar installations, commercial microgrids, and other renewable assets into mini data centers equipped with high‑end GPUs. The platform delivers on‑demand GPU capacity through a cloud‑native interface while guaranteeing transparent, usage‑based pricing with zero egress fees. Because the compute nodes draw directly from local renewable generation, they remain operational during grid outages, extreme weather, or other disruptions, providing higher availability and lower carbon impact than conventional hyperscalers. Open‑source tooling and standard APIs ensure seamless integration with existing CI/CD pipelines, orchestration frameworks, and AI platforms. Customers can provision GPU instances instantly via a web dashboard or programmatic API, with billing tracked per‑hour at published rates (e.g., $2.79 /hr for an H200 node, $1.39 /hr for an RTX PRO 6000 node).
Target Audience
Primary customers are AI/ML developers, data‑science teams, and enterprises that require high‑performance GPU compute, as well as renewable energy producers seeking to monetize excess solar generation through distributed compute services.
Features
- Solar‑fed GPU nodes (H200, RTX PRO 6000) delivering up to 141 GB VRAM, 24 vCPU, 256 GB RAM, and NVMe storage
- Microgrid integration layer that abstracts renewable energy sources as compute resources, enabling continuous operation during blackouts or extreme events
- Transparent, per‑hour pricing model with no data‑egress charges, displayed in real‑time on the provisioning portal
- Standard cloud APIs (REST, gRPC) and open‑source SDKs for Kubernetes, Terraform, and popular ML frameworks (TensorFlow, PyTorch)
- Edge‑ready deployment allowing workloads to run close to data sources, reducing latency and bandwidth costs
- Automated monitoring and fault‑tolerance stack that reports node health, power availability, and GPU utilization to a unified dashboard
- Built‑in security controls: encrypted data transit, role‑based access, and compliance‑ready logging